Feature Overview


Develop customized neural machine translation solutions using a vast repository of high quality bi-lingual training sets.

Improve machine translation quality

KantanLibrary™ is a repository of high-quality training data sets that helps improve and customize KantanMT engines quickly. Our updated KantanLibrary makes it easy for Project Managers and MT engineers to navigate through our high-quality data catalogue in various language pairs and domains. The KantanLibrary data is publicly available and is Intellectual Property Rights (IPR) cleared by our KantanMT Professional Services Team.

Engine customization - simplified!

Quality is central to each service offered to our clients. We know that quality and consistency of translation are the credentials that build your business and impress your clients. That’s why we’ve made it simple to train your MT engine to generate higher quality more consistent translations.

Everything you need to drive your localization process for global growth

Translation adds complexity to a developer’s environment. The value of automated machine translation is that it takes the complexity out of translation. After the initial setup, you will be able to focus on what counts - making your products smarter and your customer experiences more exceptional.

High quality training data sets

KantanLibrary contains over 15 billion words of training data – which can be used in engine customization. Your KantanMT engine can develop broader domain expertise and knowledge to generate higher quality translation.  To simplify this process, we’ve organised KantanLibrary into the following domains:








Customized data sets

We are constantly adding new data sets to KantanLibrary for new language pairs and new domains or industry verticals. The current list includes the following combinations. If you require a combination that is not listed, just contact our support team (support@kantanmt.com) and we’ll do our best to create it for you.

Scroll to Top